The Severities of White Matter Lesions Possibly Influence the Recurrences of Several Stroke Types Toshio Imaizumi, MD, PhD, Shigeru Inamura, MD, and Tatsufumi Nomura, MD, PhD
Background: White matter lesions (WML) have been reported to be a surrogate marker of stroke occurrences. To investigate the recurrent stroke types associated with WML, we prospectively observed recurrences in stroke patients. Methods: We prospectively analyzed the types of stroke recurrence in 807 patients (351 women, 69.8 6 12.0 years old) consecutively admitted to our hospital with strokes from April 2004 to October 2009. WML were graded using Fazekas grades. Recurrence-free rate curves were generated by the Kaplan–Meier method using the log-rank test. Results: The follow-up period was 31.6 6 22.2 (range .5-71) months. During this period, 111 strokes recurred. The incidence of all stroke recurrences in 223 patients with grades 2 and 3 was 6.0% per year, which was significantly higher than that seen in 287 patients with grade 0 (2.1% per year, P 5 .006) and 297 patients with grade 1 (4.3% per year, P ,.0001). The incidences of intracerebral hemorrhages (ICHs), lacunar infarctions, and atherothrombotic infarctions in patients with grades 2 and 3 were 1.2, 1.4, and 2.4% per year, respectively, which were higher than those seen in patients with grade 0 (.3%, P 5.004; .2%, P 5.002; and .9% per year, P 5.010) but not similar to those in patients with grade 1 (.9%, P 5.180; .9%, P 5.161; and 1.4% per year, P 5.249). There was no significant difference among grades 0, 1, 2, and 3 in incidence of stroke recurrences presenting as cardioembolic infarctions. Conclusions: The presence of high-grade WML elevated the rate of stroke recurrences presenting as ICHs, lacunar infarctions, and atherothrombotic infarctions but not as cardioembolic infarctions. Key Words: White matter lesions—microbleeds—stroke recurrence—microangiopathy—macroangiopathy. Ó 2014 by National Stroke Association
White matter lesions (WML) are commonly observed in elderly persons, even in those without a history of stroke, along with spotty high-intensity regions on fluidattenuated inversion recovery or T2-weighted magnetic resonance (MR) images. Nonetheless, it is unclear whether these findings provide useful clinical information regarding therapy.
From the Department of Neurosurgery, Kushiro City General Hospital, Kushiro, Hokkaido, Japan. Received December 3, 2013; revision received February 3, 2014; accepted February 11, 2014. Address correspondence to Toshio Imaizumi, MD, PhD, Kushiro City General Hospital, 1-12 Shunkodai, Kushiro, Hokkaido 0850822, Japan. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.02.011
WML originate from cerebral blood flow reduction associated with microangiopathies, including lipohyalinosis and amyloid angiopathy, and possibly venous diseases.1,2 WML also correlate with cerebral infarctions related to microangiopathies, particularly lacunar infarction, intracerebral hemorrhage (ICH), and cerebral microbleeds (MBs).3-7 In addition to their link to reduced cerebral blood flow,8 WML are associated with a reduction in intravascular density,9 which may lead to infarct growth in acute settings by preventing peripheral compensation during ischemic stress.10 The burden of small cerebral infarctions may be substantially underestimated by studies that define small infarcts on the basis of cavitated lacunar lesions.11 Therefore, some WML may be the consequence of small white matter infarctions in the vulnerable periventricular region in patients with aggressive cerebral small arterial diseases.2,11 Furthermore, large genome-wide
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2014: pp 1-6
1
T. IMAIZUMI ET AL.
2
association studies of the WML burden in communitybased cohorts of individuals of European descent recently identified a novel locus on chromosome 17.12 WML are an independent predictor of risk of symptomatic stroke,13 stroke recurrence,6,14 hemorrhagic transformation after thrombolysis for ischemic stroke,15 and clinical outcome after ischemic stroke.16 WML may also be surrogate markers of stroke recurrence and severity of microangiopathy. Furthermore, WML are correlated with dementia, including Alzheimer disease and cerebrovascular dementia, and a higher rate of mortality in patients with dementia17 and are an independent predictor of risk of poststroke dementia.18 In patients having Alzheimer disease, amyloid angiopathy induces WML and hippocampal atrophy.19,20 Although WML are associated with microangiopathy, little is known about the types of recurrent strokes they are associated with. To investigate this issue, we prospectively observed stroke recurrences.
Subjects and Methods Subjects From April 2004 to October 2009, we enrolled patients consecutively admitted to our hospital within 7 days of experiencing strokes (index strokes). Follow-up took place until January 2010 at the latest, and stroke recurrences were evaluated in all patients. We excluded patients who died within 2 weeks of the first admission, patients for whom we had follow-up of less than 2 weeks, and patients with unclear findings on magnetic resonance imaging (MRI) because of motion or metal artifacts. All study procedures were approved by the Ethics Committee of Kushiro City General Hospital (IRB 2004-1).
Radiological Examination More than 2 physicians with Japanese Board Certifications in both Neurosurgery and Stroke diagnosed stroke types based on radiological findings. Imaging findings were reviewed by more than 1 physician without the knowledge of clinical information or treatment assignment. Patients with neurologic deficits, including those that were transient and those lacking neuroradiological findings on computed tomography and MRI, were excluded from this study. The severity of white matter hyperintensity (WMH) or periventricular hyperintensity (PVH) on fluid-attenuated inversion recovery imaging was rated according to the Fazekas scale (WMH: grade 1, punctuate; grade 2, early confluence; and grade 3, confluent; and PVH: grade 1, caps or lining; grade 2, bands; and grade 3, irregular extension into the deep white matter).21 In this study, WMH and PVH grades of 3 on the Fazekas scale were regarded as grade 3. WMH and PVH grades less than 3 and 2 or more were regarded as grade 2. Finally, WHM and
PVH grades less than 2 and 1 or more were regarded as grade 1. No instances of WMH or PVH were defined as grade 0. Dot-like, low-intensity spots on T2*-weighted MRI with diameters less than 10 mm were defined as MBs. Lowintensity spots with diameters less than 1 mm were excluded, whereas those without focal signs and with diameters of 10 mm or more were classified as definite asymptomatic ICHs. Diagnosis and methods were described in detail in our previous articles.4,5
Variables Subjects were considered hypertensive if their blood pressure (BP) repeatedly exceeded 140/85 mm Hg or if they were taking antihypertensive medications. At each follow-up visit, antihypertensive drug therapies were titrated to achieve a target BP less than 140/85 mm Hg. Patients’ medications were recorded at the time of stroke recurrence, when repeat MRIs were performed in patients without recurrence around 1 year after the onset of index stroke, or when last medical examinations were performed with a follow-up of less than 1 year after the onset of index stroke.
Statistics We generated recurrence-free rate curves by the Kaplan–Meier method. Stroke recurrence rates were compared by the log-rank test according to the WML grade. To investigate the relationship between grade and stroke recurrence, patients were divided into subgroups based on grade. Overall frequencies of categorical variables were calculated in the form of odds ratios (ORs) and 95% confidence intervals (CIs) derived from univariate logistic regression analyses. Multivariate analyses were also performed using a regression model, with the elimination of variables that might not contribute to the model (P . .15) as identified in univariate analyses. Where applicable, 95% CIs were calculated for the estimated ORs. A P value less than .05 was considered statistically significant.
Results Of the 1095 patients consecutively enrolled during the study period, we excluded 123 who survived less than 2 weeks, 67 followed up for less than 2 weeks, and 27 with pacemakers. An additional 21 patients were also excluded as a result of unclear MRI findings because of motion or metal artifacts. Fifty one patients were excluded because of insufficient information. After these exclusions, we followed and investigated 807 patients (351 women, 69.8 6 12.0 years old) with strokes. The stroke type (index stroke) at admission was ICH in 188 patients (46 lobar and 142 deep ICHs), subarachnoid
WHITE MATTER LESIONS AND STROKE RECURRENCES
hemorrhage in 56 patients, lacunar infarction in 212 patients, atherothrombotic infarction in 208 patients, cardioembolic infarction in 119 patients, and infarction of unknown origin in 23 patients. The mean follow-up period was 31.6 6 22.2 (.5-71) months. During the follow-up period, 107 of the 807 patients were readmitted or retreated for 111 recurrent strokes, including 24 ICHs (3 lobar and 21 deep ICHs), 21 lacunar infarctions, 27 cardioembolic infarctions, 38 atherothrombotic infarctions, and 1 infarction of unknown origin. Two patients had 2 recurrent strokes each, whereas 1 patient had 3 recurrent strokes over the course of the study period. Of 807 patients, MRI demonstrated WML of grade 0 in 286 (35%) patients, grade 1 in 298 (37%) patients, grade 2 in 168 (21%) patients, and grade 3 in 55 (7%) patients (Table 1). The incidence of each stroke type did not differ significantly between patients with grades 2 and 3. Therefore, we grouped patients with grades 2 and 3 and analyzed the incidence of recurrences. Stroke recurrence-free rate curves, generated by the Kaplan–Meier method using the log-rank test, demonstrated that the incidence of all types of stroke recurrence in patients with grades 2 and 3 was 6.0% per year, which was significantly greater than that seen in patients with grade 0 (2.1% per year, P 5 .006) and grade 1 (4.3% per year, P , .0001; Fig 1). Univariate analysis demonstrated that patients of 65 years or older and with a history of ICH, WML grades 2 and 3, and MBs of 5 or more were significantly associated with an increased risk of stroke occurrence (all types; Table 1). In addition, WML grade 0 significantly correlated with a decrement in risk. The rate of stroke recurrences was significantly and independently associated with WML grades 2 and 3 (OR: 1.61, 95% CI: 1.02-2.55, P 5 .040) when adjusted for MBs of 5 or more, female gender, age 65 years or older, body mass index of 26 kg/m2 or more, hypertension, and a history of cerebral infarction and ICH, each of which resulted in P value less than .15 in the univariate analyses (model 1 in Table 2). However, although the rate of stroke recurrences was significantly associated with MBs of 5 or more, it did not correlate with WML grade 0 (OR: .70, 95% CI: .43-1.15, P 5 .163; model 2 in Table 2). The incidences of ICHs, lacunar infarctions, and atherothrombotic infarctions in patients with grades 2 and 3 were 1.2, 1.4, and 2.4% per year, respectively, which were higher than those seen in patients with grade 0 (.3%, P 5 .004; .2%, P 5 .002; and .9% per year, P 5 .010) but not those in patients with grade 1 (.9%, P 5 .180; .9%, P 5 .161; and 1.4% per year, P 5 .249; Fig 2). There was no difference between patients with grade 1 and grades 2 and 3 in terms of the incidence with which ICHs, lacunar infarctions, and atherothrombotic infarctions presented as recurrent strokes (P 5 .088, P 5 .079, and P 5 .110, respectively). Severe
3
WML contributed to stroke recurrences presenting as ICHs, lacunar infarctions, and atherothrombotic infarctions. There was no significant difference among grades 0, 1, 2, and 3 in terms of incidence of stroke recurrences presenting as cardioembolic infarctions. The incidences of deep ICHs, lacunar infarctions, and atherothrombotic infarctions presenting as stroke recurrences were statistically analyzed as dependent factors using logistic regression analyses. These showed that the incidences of recurrences presenting as deep ICHs and lacunar infarctions were significantly elevated in stroke patients with grades 2 and 3 (OR: 3.32, 95% CI: 1.38-7.99, P 5 .008; and OR: 3.09, 95% CI: 1.23-7.81, P 5 .017, respectively) and were significantly reduced in patients with grade 0 (OR: .28, 95% CI: .09-.87, P 5 .028; and OR: .28, 95% CI: .08-.99, P 5 .049, respectively), when adjusted for female gender, 65 years or older, hypertension, diabetes mellitus, and low-density lipoprotein cholesterol more than 150 mg/dL. However, the incidence of atherothrombotic infarctions did not correlate with grades 2 and 3 (OR: 1.77, 95% CI: .89-3.54, P 5 .104) or grade 0 (OR: .65, 95% CI: .29-1.45, P 5 .291).
Discussion WML and Stroke Recurrences We prospectively investigated WML as a potential clinical factor in patients with strokes and found that the incidence of stroke recurrence was significantly higher in patients with high-grade WML. It was difficult to assign definitive WML severity grades in this study because we could not investigate WML volume. Although there were ambiguous differences between grades 1 and 2 and between grades 2 and 3, the differences between grade 0 and grades 2 and 3 were clear. The apparent differences between grade 0 and grades 2 and 3 might enhance the statistical differences observed in stroke recurrences. We could not correctly calculate the volumes of WML areas on MR images in this study because we did not have absolute values of MR intensities. In future studies, we will calculate WML volume on MR images and examine the relationship between WML volume and stroke recurrences.
Stroke Recurrence Types and WML In this study, only 3 stroke recurrences presented as subcortical ICHs, and we, therefore, investigated ICHs without further subdivision into deep and lobar ICHs. WML are potentially induced by both amyloid angiopathy, associated with lobar ICHs, and hypertensive microangiopathy, associated with deep ICHs.2 Our findings may, therefore, be relevant despite our grouping of the 2 ICH types. A previous investigation revealed that amyloid angiopathy might induce WML or hippocampal atrophy in patients having Alzheimer disease.19
T. IMAIZUMI ET AL.
4
Table 1. Univariate analysis Recurrence
No recurrence
Univariate analysis
Variable
Patients, n 5 107
%
Patients, n 5 700
(%)
OR
95% CI
P value
Female Age $ 65 y BMI $ 26 kg/m2 HDL cholesterol , 40 mg/dL LDL cholesterol $ 150 mg/dL Triglyceride $ 150 mg/dL Diabetes mellitus Hypertension Hemodialysis Past IHD Past cerebral infarction Past ICH Antiplatelet drug use Warfarin use WML Gr 0 Gr 1 Gr 2-3 MBs 4 $ MBs $ 1 MBs $ 5
39 85 14 31 10 18 30 60 6 11 21 9 64 15
36.4 79.4 13.1 29 9.3 16.8 28 56.1 5.6 10.3 19.6 8.4 59.8 14
312 456 138 159 89 122 155 451 26 64 93 26 382 90
44.5 65.1 19.7 22.7 12.7 17.4 22.1 64.4 3.7 9.1 13.2 3.7 54.5 12.9
.71 2.07 .61 1.39 .71 .88 1.37 .705 1.54 1.02 1.59 2.38 1.24 1.11
.47-1.09 1.26-3.39 .34-1.11 .88-2.18 .36-1.41 .56-1.65 .87-2.17 .47-1.06 .62-3.83 .51-2.06 .94-2.69 1.08-5.23 .82-1.88 .61-1.99
.116 .004 .105 .157 .325 .958 .178 .096 .353 .949 .082 .031 .91 .74
27 36 44
25.2 33.6 41.1
259 262 179
37 37.4 25.6
.57 .85 2.03
.36-.91 .55-1.31 1.34-3.10
.018 .461 .001
28 30
26.2 28
207 118
29.6 16.9
.844 1.92
.53-1.34 1.21-3.06
.471 .006
Abbreviations: BMI, body mass index; CI, confidence interval; Gr, grade; HDL, high-density lipoprotein; ICH, intracerebral hemorrhage; IHD, ischemic heart disease; LDL, low-density lipoprotein; MB, microbleed; OR, odds ratio; WML, white matter lesion.
Higher WML grades may predict the development of ICHs and lacunar infarctions associated with microangiopathies. Atherothrombotic infarction was associated with severe WML in this study, based on the Kaplan–Meier method using the log-rank test, suggesting that WML may be exacerbated by cerebral macroangiopathy and microangiopathy. Reduction of cerebral blood flow (velocities) because of middle cerebral artery stenosis has been linked to WML.22 WML may be a predicting factor
for the severity of cerebral macroangiopathy. However, multivariate analysis revealed that WML grades 2 and 3 did not significantly increase the incidence of atherothrombotic infarction. Although there was a difference between univariate (log-rank test) and multivariate analyses, atherothrombotic infarctions might be weakly associated with severe WML. A previous investigation of patients with cerebral amyloid angiopathy demonstrated that WML occurred at a site (the periventricular white matter) distant from the site of vascular pathology (predominantly in the cortex or leptomeninges).2 These findings implicate a flowrelated mechanism for the pathogenesis of WML, rather than blood–brain barrier dysfunction with increased permeability, at least in cerebral amyloid angiopathy.2 This underscores the complicated nature of the relationship between WML and cerebral angiopathy.
Other Risk Factors
Figure 1. Stroke recurrence–free rate curves associated with the Fazekas grade. The incidence of stroke recurrences was significantly larger in patients with grades 2 and 3 than in grade 1 or 0. Abbreviation: Gr, grade.
We also examined risk factors for stroke, including treatable risk factors, in patients with recurrent strokes. Both deep and lobar MBs were associated with WML; therefore, we did not divide MBs into 2 groups, deep and lobar, according to location.3-5 MBs were shown to be associated with an increased incidence of stroke recurrence.4-6 With differences between deep and lobar MBs, our previous studies revealed that lobar MBs were
WHITE MATTER LESIONS AND STROKE RECURRENCES
5
Table 2. Multivariate analysis Model 1
Model 2
Variable
OR
95% CI
P value
WML Gr 2-3 WML Gr 0 MBs $ 5 Female Age $ 65 y BMI $ 26 kg/m2 Hypertension Past cerebral infarction Past ICH
1.61
1.02-2.55
.04
1.59 .68 1.82 .67 .6 1.21 1.85
.97-2.63 .44-1.04 1.08-3.06 .37-1.23 .39-.92 .70-2.10 .60-4.24
.068 .077 .025 .193 .02 .502 .149
OR
95% CI
P value
.7 1.73 .68 1.83 .68 .59 1.25 1.91
.43-1.15 1.05-2.81 .44-1.05 1.08-3.10 .37-1.24 .39-.91 .72-2.16 .83-4.40
.163 .03 .085 .025 .209 .017 .43 .127
Abbreviations: BMI, body mass index; CI, confidence interval; Gr, grade; HDL, high-density lipoprotein; ICH, intracerebral hemorrhage; IHD, ischemic heart disease; LDL, low-density lipoprotein; MB, microbleed; OR, odds ratio; WML, white matter lesion.
associated with the recurrence of stroke presenting as deep ICH in Japanese stroke patients.4,23 Some lobar MBs were possibly related to hypertensive microangiopathy. In this study, univariate analysis revealed that WML grade 0 were inversely associated with the incidence of stroke recurrence but not in model 2 of the multivariate analysis (Table 2) with the number of MBs 5 or more. MBs also originated from microangiopathy. MBs of 5 or more was a strong and independent factor elevating the incidence of stroke recurrences, even in the presence of WML grade 0. However, in model 2 without MBs, WML grade 0 independently and significantly reduced the incidence of stroke recurrence (data not shown).
Figure 2. Free rate curves of several stroke types associated with the severities of white matter lesions. Severity of white matter lesion influenced the incidence of stroke recurrences presented as ICH, LI, and AT but not CE. Abbreviations: AT, atherothrombotic infarction; CE, cardioembolic infarctions; ICH, intracerebral hemorrhage; LI, lacunar infarctions.
Naka et al6 demonstrated that severe WML were a predictor of stroke recurrences presenting as both ICHs and ischemic strokes, the latter including lacunar infarctions and atherothrombotic infarctions. MBs were a stronger predictor for ICH occurrences but not ischemic stroke.6 We previously demonstrated that MBs elevated the incidences of both deep ICHs and lacunar infarctions.23 These differences may be because of the variations in methodology and subject selection. Hypertension was inversely associated with the incidence of stroke recurrences. More than 60% of patients in this study had hypertension, although we maintained BP less than 140/85 mm Hg as much as possible. Nonetheless, our findings suggest that hypertension might not elevate the incidence of stroke.
T. IMAIZUMI ET AL.
6
Limitations This study has several limitations. Our analysis used a cohort study, and therefore, the possibility of bias cannot be excluded. Such bias includes a heterogeneous population, a diverse list of drug treatments, a variety of unselected risk factors, and multiple types of strokes. In addition, the follow-up period for these patients was relatively short and inconsistent. In addition, we excluded patients whose follow-up was less than 2 weeks after stroke onset. Thus, the small sample size limited the power of analysis. Larger studies are needed to confirm our observations.
References 1. Black S, Gao F, Bilbao J. Understanding white matter disease: imaging-pathological correlations in vascular cognitive impairment. Stroke 2009;40:S48-S52. 2. Smith EE. Leukoaraiosis and stroke. Stroke 2010; 41:S139-S143. 3. Poels MMF, Vernooij MW, Ikram MA, et al. Prevalence and risk factors of cerebral microbleeds. an update of the Rotterdam Scan Study. Stroke 2010;41(Suppl): S103-S106. 4. Imaizumi T, Inamura S, Kohama I, et al. Nascent lobar microbleeds and stroke recurrence. J Stroke Cerebrovasc Dis 2014;23:610-617. 5. Imaizumi T, Inamura S, Kohama I, et al. Nascent deep microbleeds and stroke recurrence. J Stroke Cerebrovasc Dis 2014;23:520-528. 6. Naka H, Nomura E, Takahashi T, et al. Combinations of the presence or absence of cerebral microbleeds and advanced white matter hyperintensity as predictors of subsequent stroke types. AJNR Am J Neuroradiol 2006; 27:830-835. 7. Mesker DJ, Poels MMF, Ikram MA, et al. Lobar distribution of cerebral microbleeds. The Rotterdam Scan Study. Arch Neurol 2011;68:656-659. 8. O’Sullivan M, Lythgoe DJ, Pereira AC, et al. Patterns of cerebral blood flow reduction in patients with ischemic leukoaraiosis. Neurology 2002;59:321-326. 9. Moody DM, Thore CR, Amstrom JA, et al. Quantification of afferent vessels shows reduced brain vascular density in subject with leukoaraiosis. Radiology 2004;233:883-890.
10. Heistad DD, Mayhan WG, Coyle P, et al. Impaired dilatation of cerebral arterioles in chronic hypertension. Blood Vessels 1990;27:258-262. 11. Potter GM, Doubal FN, Jackson CA, et al. Counting cavitating lacunes underestimates the burden of lacunar infarction. Stroke 2010;41:267-272. 12. Fornage M, Debette S, Bis JC, et al. Genome-wide association studies of cerebral white matter lesion burden: the CHARGE Consortium. Ann Neurol 2011;69:928-939. 13. Kuller LH, Longstrech WT Jr, Arnold AM, et al. Cardiovascular Health Study Collaborative Research Group. White matter hyperintensity on cranial magnetic resonance imaging. Stroke 2004;35:1821-1825. 14. Henon H, Vroylandt P, Durieu I, et al. Leukoaraiosis more than dementia is a predictor of stroke recurrence. Stroke 2003;34:2935-2940. 15. Palumbo V, Boulanger JM, Hill MD, et al, CASES Investigators. Leukoaraiosis and intracerebral hemorrhage after thrombolisis in acute stroke. Neurology 2007; 68:1020-1024. 16. Arsava EM, Rahman R, Rosand J, et al. Severity of leukoaraiosis correlates with clinical outcome after ischemic stroke. Neurology 2009;72:1403-1410. 17. Henneman WJP, Sluimer JD, Cordonnier C, et al. MRI biomarkers of vascular damage and atrophy predicting mortality in a memory clinic population. Stroke 2009; 40:492-498. 18. Degroot JC, De Leeuw FE, Oudkerk M, et al. Periventricular cerebral white matter lesions predict rate of cognitive decline. Ann Neurol 2002;52:335-341. 19. Cordonnier C, van der Flier WM. Brain microbleeds and Alzheimer’s disease: innocent observation or key player? Brain 2011;134:335-344. 20. The LADIS Study Group. 2001–2011: A decade of the LADIS (Leukoaraiosis and DISability) Study: what have we learned about white matter changes and small-vessel disease? Cerebrovasc Dis 2011;32:577-588. 21. Fazekas F, Chawluk JB, Alavi A, et al. MR signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 1987;149:351-356. 22. Wardlaw JM, Doubal FN, Eadie E, et al. Little association between intracranial arterial stenosis and lacunar stroke. Cerebrovasc Dis 2011;31:12-18. 23. Imaizumi T, Horita Y, Hashimoto Y, et al. Dot-like hemosiderin spots on T2*-weighted MRI as a predictor of stroke recurrence: a prospective study. J Neurosurg 2004;101:915-920.
Background: White matter lesions (WML) have been reported to be a surrogate marker of stroke occurrences. To investigate the recurrent stroke types associated with WML, we prospectively observed recurrences in stroke patients. Methods: We prospectively analyzed the types of stroke recurrence in 807 patients (351 women, 69.8 6 12.0 years old) consecutively admitted to our hospital with strokes from April 2004 to October 2009. WML were graded using Fazekas grades. Recurrence-free rate curves were generated by the Kaplan–Meier method using the log-rank test. Results: The follow-up period was 31.6 6 22.2 (range .5-71) months. During this period, 111 strokes recurred. The incidence of all stroke recurrences in 223 patients with grades 2 and 3 was 6.0% per year, which was significantly higher than that seen in 287 patients with grade 0 (2.1% per year, P 5 .006) and 297 patients with grade 1 (4.3% per year, P ,.0001). The incidences of intracerebral hemorrhages (ICHs), lacunar infarctions, and atherothrombotic infarctions in patients with grades 2 and 3 were 1.2, 1.4, and 2.4% per year, respectively, which were higher than those seen in patients with grade 0 (.3%, P 5.004; .2%, P 5.002; and .9% per year, P 5.010) but not similar to those in patients with grade 1 (.9%, P 5.180; .9%, P 5.161; and 1.4% per year, P 5.249). There was no significant difference among grades 0, 1, 2, and 3 in incidence of stroke recurrences presenting as cardioembolic infarctions. Conclusions: The presence of high-grade WML elevated the rate of stroke recurrences presenting as ICHs, lacunar infarctions, and atherothrombotic infarctions but not as cardioembolic infarctions. Key Words: White matter lesions—microbleeds—stroke recurrence—microangiopathy—macroangiopathy. Ó 2014 by National Stroke Association
White matter lesions (WML) are commonly observed in elderly persons, even in those without a history of stroke, along with spotty high-intensity regions on fluidattenuated inversion recovery or T2-weighted magnetic resonance (MR) images. Nonetheless, it is unclear whether these findings provide useful clinical information regarding therapy.
From the Department of Neurosurgery, Kushiro City General Hospital, Kushiro, Hokkaido, Japan. Received December 3, 2013; revision received February 3, 2014; accepted February 11, 2014. Address correspondence to Toshio Imaizumi, MD, PhD, Kushiro City General Hospital, 1-12 Shunkodai, Kushiro, Hokkaido 0850822, Japan. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.02.011
WML originate from cerebral blood flow reduction associated with microangiopathies, including lipohyalinosis and amyloid angiopathy, and possibly venous diseases.1,2 WML also correlate with cerebral infarctions related to microangiopathies, particularly lacunar infarction, intracerebral hemorrhage (ICH), and cerebral microbleeds (MBs).3-7 In addition to their link to reduced cerebral blood flow,8 WML are associated with a reduction in intravascular density,9 which may lead to infarct growth in acute settings by preventing peripheral compensation during ischemic stress.10 The burden of small cerebral infarctions may be substantially underestimated by studies that define small infarcts on the basis of cavitated lacunar lesions.11 Therefore, some WML may be the consequence of small white matter infarctions in the vulnerable periventricular region in patients with aggressive cerebral small arterial diseases.2,11 Furthermore, large genome-wide
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2014: pp 1-6
1
T. IMAIZUMI ET AL.
2
association studies of the WML burden in communitybased cohorts of individuals of European descent recently identified a novel locus on chromosome 17.12 WML are an independent predictor of risk of symptomatic stroke,13 stroke recurrence,6,14 hemorrhagic transformation after thrombolysis for ischemic stroke,15 and clinical outcome after ischemic stroke.16 WML may also be surrogate markers of stroke recurrence and severity of microangiopathy. Furthermore, WML are correlated with dementia, including Alzheimer disease and cerebrovascular dementia, and a higher rate of mortality in patients with dementia17 and are an independent predictor of risk of poststroke dementia.18 In patients having Alzheimer disease, amyloid angiopathy induces WML and hippocampal atrophy.19,20 Although WML are associated with microangiopathy, little is known about the types of recurrent strokes they are associated with. To investigate this issue, we prospectively observed stroke recurrences.
Subjects and Methods Subjects From April 2004 to October 2009, we enrolled patients consecutively admitted to our hospital within 7 days of experiencing strokes (index strokes). Follow-up took place until January 2010 at the latest, and stroke recurrences were evaluated in all patients. We excluded patients who died within 2 weeks of the first admission, patients for whom we had follow-up of less than 2 weeks, and patients with unclear findings on magnetic resonance imaging (MRI) because of motion or metal artifacts. All study procedures were approved by the Ethics Committee of Kushiro City General Hospital (IRB 2004-1).
Radiological Examination More than 2 physicians with Japanese Board Certifications in both Neurosurgery and Stroke diagnosed stroke types based on radiological findings. Imaging findings were reviewed by more than 1 physician without the knowledge of clinical information or treatment assignment. Patients with neurologic deficits, including those that were transient and those lacking neuroradiological findings on computed tomography and MRI, were excluded from this study. The severity of white matter hyperintensity (WMH) or periventricular hyperintensity (PVH) on fluid-attenuated inversion recovery imaging was rated according to the Fazekas scale (WMH: grade 1, punctuate; grade 2, early confluence; and grade 3, confluent; and PVH: grade 1, caps or lining; grade 2, bands; and grade 3, irregular extension into the deep white matter).21 In this study, WMH and PVH grades of 3 on the Fazekas scale were regarded as grade 3. WMH and PVH grades less than 3 and 2 or more were regarded as grade 2. Finally, WHM and
PVH grades less than 2 and 1 or more were regarded as grade 1. No instances of WMH or PVH were defined as grade 0. Dot-like, low-intensity spots on T2*-weighted MRI with diameters less than 10 mm were defined as MBs. Lowintensity spots with diameters less than 1 mm were excluded, whereas those without focal signs and with diameters of 10 mm or more were classified as definite asymptomatic ICHs. Diagnosis and methods were described in detail in our previous articles.4,5
Variables Subjects were considered hypertensive if their blood pressure (BP) repeatedly exceeded 140/85 mm Hg or if they were taking antihypertensive medications. At each follow-up visit, antihypertensive drug therapies were titrated to achieve a target BP less than 140/85 mm Hg. Patients’ medications were recorded at the time of stroke recurrence, when repeat MRIs were performed in patients without recurrence around 1 year after the onset of index stroke, or when last medical examinations were performed with a follow-up of less than 1 year after the onset of index stroke.
Statistics We generated recurrence-free rate curves by the Kaplan–Meier method. Stroke recurrence rates were compared by the log-rank test according to the WML grade. To investigate the relationship between grade and stroke recurrence, patients were divided into subgroups based on grade. Overall frequencies of categorical variables were calculated in the form of odds ratios (ORs) and 95% confidence intervals (CIs) derived from univariate logistic regression analyses. Multivariate analyses were also performed using a regression model, with the elimination of variables that might not contribute to the model (P . .15) as identified in univariate analyses. Where applicable, 95% CIs were calculated for the estimated ORs. A P value less than .05 was considered statistically significant.
Results Of the 1095 patients consecutively enrolled during the study period, we excluded 123 who survived less than 2 weeks, 67 followed up for less than 2 weeks, and 27 with pacemakers. An additional 21 patients were also excluded as a result of unclear MRI findings because of motion or metal artifacts. Fifty one patients were excluded because of insufficient information. After these exclusions, we followed and investigated 807 patients (351 women, 69.8 6 12.0 years old) with strokes. The stroke type (index stroke) at admission was ICH in 188 patients (46 lobar and 142 deep ICHs), subarachnoid
WHITE MATTER LESIONS AND STROKE RECURRENCES
hemorrhage in 56 patients, lacunar infarction in 212 patients, atherothrombotic infarction in 208 patients, cardioembolic infarction in 119 patients, and infarction of unknown origin in 23 patients. The mean follow-up period was 31.6 6 22.2 (.5-71) months. During the follow-up period, 107 of the 807 patients were readmitted or retreated for 111 recurrent strokes, including 24 ICHs (3 lobar and 21 deep ICHs), 21 lacunar infarctions, 27 cardioembolic infarctions, 38 atherothrombotic infarctions, and 1 infarction of unknown origin. Two patients had 2 recurrent strokes each, whereas 1 patient had 3 recurrent strokes over the course of the study period. Of 807 patients, MRI demonstrated WML of grade 0 in 286 (35%) patients, grade 1 in 298 (37%) patients, grade 2 in 168 (21%) patients, and grade 3 in 55 (7%) patients (Table 1). The incidence of each stroke type did not differ significantly between patients with grades 2 and 3. Therefore, we grouped patients with grades 2 and 3 and analyzed the incidence of recurrences. Stroke recurrence-free rate curves, generated by the Kaplan–Meier method using the log-rank test, demonstrated that the incidence of all types of stroke recurrence in patients with grades 2 and 3 was 6.0% per year, which was significantly greater than that seen in patients with grade 0 (2.1% per year, P 5 .006) and grade 1 (4.3% per year, P , .0001; Fig 1). Univariate analysis demonstrated that patients of 65 years or older and with a history of ICH, WML grades 2 and 3, and MBs of 5 or more were significantly associated with an increased risk of stroke occurrence (all types; Table 1). In addition, WML grade 0 significantly correlated with a decrement in risk. The rate of stroke recurrences was significantly and independently associated with WML grades 2 and 3 (OR: 1.61, 95% CI: 1.02-2.55, P 5 .040) when adjusted for MBs of 5 or more, female gender, age 65 years or older, body mass index of 26 kg/m2 or more, hypertension, and a history of cerebral infarction and ICH, each of which resulted in P value less than .15 in the univariate analyses (model 1 in Table 2). However, although the rate of stroke recurrences was significantly associated with MBs of 5 or more, it did not correlate with WML grade 0 (OR: .70, 95% CI: .43-1.15, P 5 .163; model 2 in Table 2). The incidences of ICHs, lacunar infarctions, and atherothrombotic infarctions in patients with grades 2 and 3 were 1.2, 1.4, and 2.4% per year, respectively, which were higher than those seen in patients with grade 0 (.3%, P 5 .004; .2%, P 5 .002; and .9% per year, P 5 .010) but not those in patients with grade 1 (.9%, P 5 .180; .9%, P 5 .161; and 1.4% per year, P 5 .249; Fig 2). There was no difference between patients with grade 1 and grades 2 and 3 in terms of the incidence with which ICHs, lacunar infarctions, and atherothrombotic infarctions presented as recurrent strokes (P 5 .088, P 5 .079, and P 5 .110, respectively). Severe
3
WML contributed to stroke recurrences presenting as ICHs, lacunar infarctions, and atherothrombotic infarctions. There was no significant difference among grades 0, 1, 2, and 3 in terms of incidence of stroke recurrences presenting as cardioembolic infarctions. The incidences of deep ICHs, lacunar infarctions, and atherothrombotic infarctions presenting as stroke recurrences were statistically analyzed as dependent factors using logistic regression analyses. These showed that the incidences of recurrences presenting as deep ICHs and lacunar infarctions were significantly elevated in stroke patients with grades 2 and 3 (OR: 3.32, 95% CI: 1.38-7.99, P 5 .008; and OR: 3.09, 95% CI: 1.23-7.81, P 5 .017, respectively) and were significantly reduced in patients with grade 0 (OR: .28, 95% CI: .09-.87, P 5 .028; and OR: .28, 95% CI: .08-.99, P 5 .049, respectively), when adjusted for female gender, 65 years or older, hypertension, diabetes mellitus, and low-density lipoprotein cholesterol more than 150 mg/dL. However, the incidence of atherothrombotic infarctions did not correlate with grades 2 and 3 (OR: 1.77, 95% CI: .89-3.54, P 5 .104) or grade 0 (OR: .65, 95% CI: .29-1.45, P 5 .291).
Discussion WML and Stroke Recurrences We prospectively investigated WML as a potential clinical factor in patients with strokes and found that the incidence of stroke recurrence was significantly higher in patients with high-grade WML. It was difficult to assign definitive WML severity grades in this study because we could not investigate WML volume. Although there were ambiguous differences between grades 1 and 2 and between grades 2 and 3, the differences between grade 0 and grades 2 and 3 were clear. The apparent differences between grade 0 and grades 2 and 3 might enhance the statistical differences observed in stroke recurrences. We could not correctly calculate the volumes of WML areas on MR images in this study because we did not have absolute values of MR intensities. In future studies, we will calculate WML volume on MR images and examine the relationship between WML volume and stroke recurrences.
Stroke Recurrence Types and WML In this study, only 3 stroke recurrences presented as subcortical ICHs, and we, therefore, investigated ICHs without further subdivision into deep and lobar ICHs. WML are potentially induced by both amyloid angiopathy, associated with lobar ICHs, and hypertensive microangiopathy, associated with deep ICHs.2 Our findings may, therefore, be relevant despite our grouping of the 2 ICH types. A previous investigation revealed that amyloid angiopathy might induce WML or hippocampal atrophy in patients having Alzheimer disease.19
T. IMAIZUMI ET AL.
4
Table 1. Univariate analysis Recurrence
No recurrence
Univariate analysis
Variable
Patients, n 5 107
%
Patients, n 5 700
(%)
OR
95% CI
P value
Female Age $ 65 y BMI $ 26 kg/m2 HDL cholesterol , 40 mg/dL LDL cholesterol $ 150 mg/dL Triglyceride $ 150 mg/dL Diabetes mellitus Hypertension Hemodialysis Past IHD Past cerebral infarction Past ICH Antiplatelet drug use Warfarin use WML Gr 0 Gr 1 Gr 2-3 MBs 4 $ MBs $ 1 MBs $ 5
39 85 14 31 10 18 30 60 6 11 21 9 64 15
36.4 79.4 13.1 29 9.3 16.8 28 56.1 5.6 10.3 19.6 8.4 59.8 14
312 456 138 159 89 122 155 451 26 64 93 26 382 90
44.5 65.1 19.7 22.7 12.7 17.4 22.1 64.4 3.7 9.1 13.2 3.7 54.5 12.9
.71 2.07 .61 1.39 .71 .88 1.37 .705 1.54 1.02 1.59 2.38 1.24 1.11
.47-1.09 1.26-3.39 .34-1.11 .88-2.18 .36-1.41 .56-1.65 .87-2.17 .47-1.06 .62-3.83 .51-2.06 .94-2.69 1.08-5.23 .82-1.88 .61-1.99
.116 .004 .105 .157 .325 .958 .178 .096 .353 .949 .082 .031 .91 .74
27 36 44
25.2 33.6 41.1
259 262 179
37 37.4 25.6
.57 .85 2.03
.36-.91 .55-1.31 1.34-3.10
.018 .461 .001
28 30
26.2 28
207 118
29.6 16.9
.844 1.92
.53-1.34 1.21-3.06
.471 .006
Abbreviations: BMI, body mass index; CI, confidence interval; Gr, grade; HDL, high-density lipoprotein; ICH, intracerebral hemorrhage; IHD, ischemic heart disease; LDL, low-density lipoprotein; MB, microbleed; OR, odds ratio; WML, white matter lesion.
Higher WML grades may predict the development of ICHs and lacunar infarctions associated with microangiopathies. Atherothrombotic infarction was associated with severe WML in this study, based on the Kaplan–Meier method using the log-rank test, suggesting that WML may be exacerbated by cerebral macroangiopathy and microangiopathy. Reduction of cerebral blood flow (velocities) because of middle cerebral artery stenosis has been linked to WML.22 WML may be a predicting factor
for the severity of cerebral macroangiopathy. However, multivariate analysis revealed that WML grades 2 and 3 did not significantly increase the incidence of atherothrombotic infarction. Although there was a difference between univariate (log-rank test) and multivariate analyses, atherothrombotic infarctions might be weakly associated with severe WML. A previous investigation of patients with cerebral amyloid angiopathy demonstrated that WML occurred at a site (the periventricular white matter) distant from the site of vascular pathology (predominantly in the cortex or leptomeninges).2 These findings implicate a flowrelated mechanism for the pathogenesis of WML, rather than blood–brain barrier dysfunction with increased permeability, at least in cerebral amyloid angiopathy.2 This underscores the complicated nature of the relationship between WML and cerebral angiopathy.
Other Risk Factors
Figure 1. Stroke recurrence–free rate curves associated with the Fazekas grade. The incidence of stroke recurrences was significantly larger in patients with grades 2 and 3 than in grade 1 or 0. Abbreviation: Gr, grade.
We also examined risk factors for stroke, including treatable risk factors, in patients with recurrent strokes. Both deep and lobar MBs were associated with WML; therefore, we did not divide MBs into 2 groups, deep and lobar, according to location.3-5 MBs were shown to be associated with an increased incidence of stroke recurrence.4-6 With differences between deep and lobar MBs, our previous studies revealed that lobar MBs were
WHITE MATTER LESIONS AND STROKE RECURRENCES
5
Table 2. Multivariate analysis Model 1
Model 2
Variable
OR
95% CI
P value
WML Gr 2-3 WML Gr 0 MBs $ 5 Female Age $ 65 y BMI $ 26 kg/m2 Hypertension Past cerebral infarction Past ICH
1.61
1.02-2.55
.04
1.59 .68 1.82 .67 .6 1.21 1.85
.97-2.63 .44-1.04 1.08-3.06 .37-1.23 .39-.92 .70-2.10 .60-4.24
.068 .077 .025 .193 .02 .502 .149
OR
95% CI
P value
.7 1.73 .68 1.83 .68 .59 1.25 1.91
.43-1.15 1.05-2.81 .44-1.05 1.08-3.10 .37-1.24 .39-.91 .72-2.16 .83-4.40
.163 .03 .085 .025 .209 .017 .43 .127
Abbreviations: BMI, body mass index; CI, confidence interval; Gr, grade; HDL, high-density lipoprotein; ICH, intracerebral hemorrhage; IHD, ischemic heart disease; LDL, low-density lipoprotein; MB, microbleed; OR, odds ratio; WML, white matter lesion.
associated with the recurrence of stroke presenting as deep ICH in Japanese stroke patients.4,23 Some lobar MBs were possibly related to hypertensive microangiopathy. In this study, univariate analysis revealed that WML grade 0 were inversely associated with the incidence of stroke recurrence but not in model 2 of the multivariate analysis (Table 2) with the number of MBs 5 or more. MBs also originated from microangiopathy. MBs of 5 or more was a strong and independent factor elevating the incidence of stroke recurrences, even in the presence of WML grade 0. However, in model 2 without MBs, WML grade 0 independently and significantly reduced the incidence of stroke recurrence (data not shown).
Figure 2. Free rate curves of several stroke types associated with the severities of white matter lesions. Severity of white matter lesion influenced the incidence of stroke recurrences presented as ICH, LI, and AT but not CE. Abbreviations: AT, atherothrombotic infarction; CE, cardioembolic infarctions; ICH, intracerebral hemorrhage; LI, lacunar infarctions.
Naka et al6 demonstrated that severe WML were a predictor of stroke recurrences presenting as both ICHs and ischemic strokes, the latter including lacunar infarctions and atherothrombotic infarctions. MBs were a stronger predictor for ICH occurrences but not ischemic stroke.6 We previously demonstrated that MBs elevated the incidences of both deep ICHs and lacunar infarctions.23 These differences may be because of the variations in methodology and subject selection. Hypertension was inversely associated with the incidence of stroke recurrences. More than 60% of patients in this study had hypertension, although we maintained BP less than 140/85 mm Hg as much as possible. Nonetheless, our findings suggest that hypertension might not elevate the incidence of stroke.
T. IMAIZUMI ET AL.
6
Limitations This study has several limitations. Our analysis used a cohort study, and therefore, the possibility of bias cannot be excluded. Such bias includes a heterogeneous population, a diverse list of drug treatments, a variety of unselected risk factors, and multiple types of strokes. In addition, the follow-up period for these patients was relatively short and inconsistent. In addition, we excluded patients whose follow-up was less than 2 weeks after stroke onset. Thus, the small sample size limited the power of analysis. Larger studies are needed to confirm our observations.
References 1. Black S, Gao F, Bilbao J. Understanding white matter disease: imaging-pathological correlations in vascular cognitive impairment. Stroke 2009;40:S48-S52. 2. Smith EE. Leukoaraiosis and stroke. Stroke 2010; 41:S139-S143. 3. Poels MMF, Vernooij MW, Ikram MA, et al. Prevalence and risk factors of cerebral microbleeds. an update of the Rotterdam Scan Study. Stroke 2010;41(Suppl): S103-S106. 4. Imaizumi T, Inamura S, Kohama I, et al. Nascent lobar microbleeds and stroke recurrence. J Stroke Cerebrovasc Dis 2014;23:610-617. 5. Imaizumi T, Inamura S, Kohama I, et al. Nascent deep microbleeds and stroke recurrence. J Stroke Cerebrovasc Dis 2014;23:520-528. 6. Naka H, Nomura E, Takahashi T, et al. Combinations of the presence or absence of cerebral microbleeds and advanced white matter hyperintensity as predictors of subsequent stroke types. AJNR Am J Neuroradiol 2006; 27:830-835. 7. Mesker DJ, Poels MMF, Ikram MA, et al. Lobar distribution of cerebral microbleeds. The Rotterdam Scan Study. Arch Neurol 2011;68:656-659. 8. O’Sullivan M, Lythgoe DJ, Pereira AC, et al. Patterns of cerebral blood flow reduction in patients with ischemic leukoaraiosis. Neurology 2002;59:321-326. 9. Moody DM, Thore CR, Amstrom JA, et al. Quantification of afferent vessels shows reduced brain vascular density in subject with leukoaraiosis. Radiology 2004;233:883-890.
10. Heistad DD, Mayhan WG, Coyle P, et al. Impaired dilatation of cerebral arterioles in chronic hypertension. Blood Vessels 1990;27:258-262. 11. Potter GM, Doubal FN, Jackson CA, et al. Counting cavitating lacunes underestimates the burden of lacunar infarction. Stroke 2010;41:267-272. 12. Fornage M, Debette S, Bis JC, et al. Genome-wide association studies of cerebral white matter lesion burden: the CHARGE Consortium. Ann Neurol 2011;69:928-939. 13. Kuller LH, Longstrech WT Jr, Arnold AM, et al. Cardiovascular Health Study Collaborative Research Group. White matter hyperintensity on cranial magnetic resonance imaging. Stroke 2004;35:1821-1825. 14. Henon H, Vroylandt P, Durieu I, et al. Leukoaraiosis more than dementia is a predictor of stroke recurrence. Stroke 2003;34:2935-2940. 15. Palumbo V, Boulanger JM, Hill MD, et al, CASES Investigators. Leukoaraiosis and intracerebral hemorrhage after thrombolisis in acute stroke. Neurology 2007; 68:1020-1024. 16. Arsava EM, Rahman R, Rosand J, et al. Severity of leukoaraiosis correlates with clinical outcome after ischemic stroke. Neurology 2009;72:1403-1410. 17. Henneman WJP, Sluimer JD, Cordonnier C, et al. MRI biomarkers of vascular damage and atrophy predicting mortality in a memory clinic population. Stroke 2009; 40:492-498. 18. Degroot JC, De Leeuw FE, Oudkerk M, et al. Periventricular cerebral white matter lesions predict rate of cognitive decline. Ann Neurol 2002;52:335-341. 19. Cordonnier C, van der Flier WM. Brain microbleeds and Alzheimer’s disease: innocent observation or key player? Brain 2011;134:335-344. 20. The LADIS Study Group. 2001–2011: A decade of the LADIS (Leukoaraiosis and DISability) Study: what have we learned about white matter changes and small-vessel disease? Cerebrovasc Dis 2011;32:577-588. 21. Fazekas F, Chawluk JB, Alavi A, et al. MR signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 1987;149:351-356. 22. Wardlaw JM, Doubal FN, Eadie E, et al. Little association between intracranial arterial stenosis and lacunar stroke. Cerebrovasc Dis 2011;31:12-18. 23. Imaizumi T, Horita Y, Hashimoto Y, et al. Dot-like hemosiderin spots on T2*-weighted MRI as a predictor of stroke recurrence: a prospective study. J Neurosurg 2004;101:915-920.
